Thicken Polysrf

Hey all,

I’ve got this polysrface that I want to give thickness to- what is the best way to do that? Offsetsrf is generating an open polysrf. Might it be a curvature continuity thing? Are my surfaces not properly joined? The original polysrf only has naked edges around the edges, but the thickened model has naked edges in the middle.

Thickened model:

160516 Temp.3dm (230.0 KB)

Hi Lawrence - the problem area looks like it is in that pointy shape, the triangle - that is kind of bubbly on the original - very high curvature near the point, so offsetting is mot liking it much. You need to make that surface more cleanly.

It’s a little unclear how you want all that to come together - here is a quick and dirty ‘fix’ with Patch to make the offset work OK, but I’d say you need to resolve the design around that whole area. The marked edge was matched up for curvature as well as the Patch fix.

160516 Temp_PG.3dm (211.4 KB)


Thanks @pascal ,

Your lower patch with the 5 edges, how did you make that? It seems to have a clean UV and then a trim on its bottom corner.

In the case of offsetsrf, the thickened edges (in my case 1.5mm) seems to be generated by extruding perpendicular to the surface. If I want more control of how those edges look, would I offset the surface (non-solid) and then try to sweep2 between the two polysrfs? Are there any other commands that might thicken a surface?

Hi Lawrence - I trimmed back the squiggly surface and filled in the area with the Patch command - not ideal, but it does avoid that pointy structure that you had in the triangular piece - I’d avoid that. But as I say, it looks like overall, the thing needs a bit of attention to see what the design intent really is, and (probably) simpler construction.


The challenge for me is that I don’t always know what I want the final thing to look like before I build it, so I can’t anticipate how the surfaces will blend.

I have a question about blendsrf. It seems that sometimes the control points are not pointing the direction I want. Often I want them to point towards the other surface’s control point, but when I rotate the control points, they don’t always want to snap to the corner for some reason. Is there an easy way to get the control points of both sides to point towards each other?

Hi Lawrence- I just line the points up visually, but in this case it is pretty hard to do, on the lower set.


Blendsrf input controls are temperamental. There is a hotkey toggle to help control them a bit. The help file explains better than I can.

Hey @pascal,

I have another polysrf that is giving me problems. I have a feeling that it would better to rebuild the thing before thickening it, but this might be a good learning opportunity. No matter what I do the hole in the model does not want to close. I’ve tried sweep2 and patch and neither seems to work. For some reason the edges just don’t want to match up. What is your strategy for a situation like this?

160517 Temp.3dm (942.5 KB)

If you were to remodel a surface like this how would you do it? This is my intent:

Hi Lawrence - I’d put the singularity at the other end of that roundy surface - hold on a second, I’ll make an example.

160517 Temp_PG.3dm (262.8 KB)


How did you make that surface? Was that a sweep1 with sections that converge at the singularity?

Hi Lawrence - [quote=“lawrenceyy, post:10, topic:32523”]
Was that a sweep1 with sections that converge at the singularity?

That would work as well… because I like to keep things simple and all three shape curves have the same structure, I lofted them first, then MatchSrf for tangency back to the straight surfaces, then for Curvature around the curved edge, with a dose or two of InsertKnot > Automatic before that last operation to give it enough points to line up ok along the edge.
The main thing is that singularity where all the points converge is at a much more convenient location.

@lawrenceyy - for some reason, you have a pile of coincident or very nearly coincident points in the long ‘side’ surfaces, where they start to bend in (in plan). I’d clean that up.


I’ve been using matchsrf more recently, but I am not exactly sure what it does to the surface. Does it just bend the surface edge so that it matches the target edge while coercing it for tangency and continuity? If my surface is four-sided, are the other three edges safe during this operation?

What exactly does insertknot do for us in this situation? Insertknot looks a lot like rebuild to me.

Hi Lawrence -[quote=“lawrenceyy, post:12, topic:32523”]
If my surface is four-sided, are the other three edges safe during this operation?

No - That is, the opposite edge is safe, depending on the ‘Preserve other end’ setting in the dialog but not the side edges.[quote=“lawrenceyy, post:12, topic:32523”]
What exactly does insertknot do for us in this situation? Insertknot looks a lot like rebuild to me.
It is not really like Rebuild. Inserting knots increases the complexity of a surface or curve- it does not change the 3d shape of the object at all. Rebuild, except in some special cases (like planes) always changes the shape of the object, either a little or a lot. Rebuild results in a uniform object (knot spacing) and InsertKnot rarely does, quite, but with Automatic it does if the input is uniform - Automatic splits every existing span evenly into two. I did not want to change the shape or the structure in the radial direction at all except by MatchSrf, but I needed enough points to be able to match to the target edge reasonably closely, so InsertKnot.



I tried over and over but I can’t seem to get my lofted surface to join and offsetsrf properly. I followed your instructions, and thanks for the explanation on insertknot. Can you screencap your process?

Is there anyway to tell if it will offsetsrf properly before you join and actually test out the command?

160517 Temp_2.3dm (167.9 KB)

Hi Lawrence - I guess I only matched for tangency at the front edge - the sides are only tangent to the bottom.


Thanks @pascal. I followed your instructions exactly and it worked.

I didn’t realize that you lofted from the flat piece to the middle curve and then back to the flat piece. I first extracted a few isocurves (u shape same as bendy piece) from the old surface and then lofted from the bendy piece to the isocurves and then to the center point. My problems after offsetsrf stemmed from the singularity. I wonder why your singularity did not cause problems but mine did…

Seems kind of strange that you can do similar things to get similar resultant surfaces, but only one process’s surface is actually offsetsrf-able.

Hey @pascal,

I went back yesterday and remodeled the original surface so that it has better curvature. I still have that singularity, but it’s causing me issues when I offsetsrf. I eventually remodeled the problem surfaces (after the offset) and got a closed polysrf, but I’d like to know how I should handle the singularity in the future. Would you ever trim the singularity and try to patch it with another small surface?

160518 Temp_1.3dm (339.4 KB)

Hi Lawrence - the two halves are not tangent along the centerline - that will mess things up.

and the singularity is not very smooth:

160518 Temp_1_pg.3dm (137.7 KB)


I don’t get it. I did matchsrf on every edge, but it would never fix the singularity. Which command did you use to fix the singularity? Zebra seems to pinch at the singularity no matter what for me. Is this a tolerance thing?

I noticed yesterday when I offsetsrf with a tolerance of .001, it would produce a open polysrf, but when the tolerance was .01 it would create a closed surface. Why do some commands not operate on the tolerances set in the options menu?

Nice hack, but “best practice” modeling would be to avoid the singularity all together, which shouldn’t be too difficult in this case.